Recommend
2 
 Thumb up
 Hide
10 Posts

Backgammon» Forums » General

Subject: What's a good win rate? rss

Your Tags: Add tags
Popular Tags: [View All]
Matt Thrower
United Kingdom
Bath
Somerset
flag msg tools
badge
Avatar
mbmbmbmbmb
I have a backgammon program on my smartphone which features an embarrassingly bad AI. It also has a handy stats function which tracks your win rate against the dreadful computer player. A "win" counts as the first to three points in a set of matches using the doubling cube - so it can take as many as five games or as few as one to get a "win".

Given that backgammon obviously involves a random element, whatever win rate you manage is not entirely a reflection of your skill. Even the lowliest player should be able to beat a world champion a small proportion of the time. So the question is, allowing for random variation, what's a "good" win rate for backgammon? I'm managing an average of 75% against my poor computer AI - what does that say about my skill (or lack thereof) at Backgammon?
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Benedikt Rosenau
Germany
Göttingen
flag msg tools
designer
mbmbmbmbmb
MattDP wrote:
I'm managing an average of 75% against my poor computer AI - what does that say about my skill (or lack thereof) at Backgammon?

Assuming your 75% refers to single games, not to three point matches:

I heard that the best players win around 82% of the time against a not fully specified class of players. I assume this class is made up of the average player who knows some basics such as what a prime is. So, there are players even below them. Given that you win 75% of the time and that you are probably not a top class player, I assume the program belongs to the class below the reference class for the 82%.

In other words, it is pretty bad.
2 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Matt Thrower
United Kingdom
Bath
Somerset
flag msg tools
badge
Avatar
mbmbmbmbmb
Zickzack wrote:
Assuming your 75% refers to single games, not to three point matches:.


Actually no, it refers to the 3-point matches.

But that 82% stat is interesting, thanks.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Todd Redden
United States
Manchester
Connecticut
flag msg tools
Nut'n honey
badge
"One of the key problems today is that politics is such a disgrace, good people don't go into government." - Donald Trump
Avatar
mbmbmbmbmb
It's a hard question to answer. Backgammon more than perhaps any other software is written to cheat. That's not a can of worms, in fact at least a few programmers have admitted their AI favors certain die rolls for itself, and does other things that improve the program's chances of winning. The very best Backgammon software, of course, uses completely randomizing die rolls. And, many programs offer settings that effect the programs playing style (attacking, defending, leaving stones uncovered, etc.)

As for what to expect against a player (or software) of equal skills is anybody's guess. I once won against an equal player 10 games in a row, and the next 10 games in a row were one won by him. You'd have to analyze statistically in order to come up with a certain win rate against whatever software AI you're playing. There could not be a possible standard for comparison.
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Was George Orwell an Optimist?
United States
Corvallis
Oregon
flag msg tools
Tete Montoliu - Tootie's Tempo
badge
Avatar
mbmbmbmbmb
82% sounds like an awfully specific number to reach for 'unspecified' opponents. Your win rate only has two inputs, your ability and the opponent's ability. If you're trying to gauge your own ability, the strength of the opponent is the only input that matters.

To find out how good your win rate is against a specific AI, I think you would need to determine your own over a statistically significant number of games, and compare with the win rate of others who have done the same. Anything else is smoke and mirrors.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Tim Koppang
United States
Westmont
Illinois
flag msg tools
designer
publisher
"It's a magical world, Hobbes, ol' buddy..."
badge
"For the listener, who listens in the snow, and, nothing himself, beholds nothing that is not there and the nothing that is." -- Wallace Stevens
Avatar
mbmbmbmbmb
tmredden wrote:
It's a hard question to answer. Backgammon more than perhaps any other software is written to cheat.

I'm not saying that certain programs do not cheat, but I think that Backgammon programs, more than most others, are often suspected of cheating. Whether or not they actually do is an entirely different question.

That said, the best way to find out your own skill level is to determine what your ranking is as compared to the computer program's ranking (if it could be determined).
1 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Tyler McLaughlin
Canada
Medicine Hat
Alberta
flag msg tools
Talk
badge
Words
Avatar
mbmbmbmbmb
Around 80% is a better idea.

The chump wins around 1 out of 5 games against a shark.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Benedikt Rosenau
Germany
Göttingen
flag msg tools
designer
mbmbmbmbmb
Sphere wrote:
82% sounds like an awfully specific number to reach for 'unspecified' opponents.

One group is world class players, and the other, somewhat less clearly, means relative beginners.
 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Richard Morris
Scotland
Harrogate
North Yorkshire
flag msg tools
designer
Join the BGG Folding @Home Team !!
badge
This user had more :gg: than sense
Avatar
mbmbmbmbmb
Zickzack wrote:
Sphere wrote:
82% sounds like an awfully specific number to reach for 'unspecified' opponents.

One group is world class players, and the other, somewhat less clearly, means relative beginners.


Relative beginners? On the definition I would use of relative, I find that very hard to accept. Sure there is plenty of luck with the dice rolls, but a World class player knows how best to play to minimise the effect of bad rolls, and maximise the effect of good ones, when to play aggressively, and when not to, and so on. Whilst I can see a 'relative beginner' getting lucky the multiple times in a game necessary to win when they have no idea of strategy, I cannot see that it would happen in nearly 1 game in 5. I could understand it more if the lesser player had at least a reasonable idea of strategy - ie past the beginner phase.

And, of course, results in single games when using the doubling cube are not terribly meaningful, since a good player will be playing the 'meta game' of the series, more than the single game. Without that, you have to play out every game to its conclusion, which means that the luck of the dice can have more effect (and you might as well be playing ludo). When I used to play online, I recall quite a few games that I lost (and others that I won) because the doubling cube was not in play, which would almost certainly have ended much earlier with the opposite result with a refused double.
2 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Todd Redden
United States
Manchester
Connecticut
flag msg tools
Nut'n honey
badge
"One of the key problems today is that politics is such a disgrace, good people don't go into government." - Donald Trump
Avatar
mbmbmbmbmb
tckoppang wrote:
tmredden wrote:
It's a hard question to answer. Backgammon more than perhaps any other software is written to cheat.

I'm not saying that certain programs do not cheat, but I think that Backgammon programs, more than most others, are often suspected of cheating. Whether or not they actually do is an entirely different question.

That said, the best way to find out your own skill level is to determine what your ranking is as compared to the computer program's ranking (if it could be determined).

Admittedly, a lot of backgammon programmers have been charged with writing cheating into their apps, in most cases probably unfairly. What happens is players do statistical analysis on the program's die rolls and discover an excess of doubles than would normally be caused by truly random die rolls. It could be the programmer is using a bad seed or poorly programmed method of producing random results or, in some cases admittedly by the programmer, the program's superior results were intentional. Most backgammon programmers so proclaimed have made their code available for inspection. It really is disheartening when a program repeatedly rolls exactly the numbers needed to take out your one open stone and proceed to blocks you off mercilessly when the chances highly favored other results.
2 
 Thumb up
 tip
 Hide
  • [+] Dice rolls
Front Page | Welcome | Contact | Privacy Policy | Terms of Service | Advertise | Support BGG | Feeds RSS
Geekdo, BoardGameGeek, the Geekdo logo, and the BoardGameGeek logo are trademarks of BoardGameGeek, LLC.